User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.6) Gecko/20050225 Firefox/1.0.1 StumbleUpon/1.9992 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.6) Gecko/20050225 Firefox/1.0.1 StumbleUpon/1.9992 I will attach a test case in a moment. This is an HTML which consists of: 1) a small CSS declaration at the top 2) a big HTML comment with loads of x's in it 3) 10 identical(!) copies of an <a> element with two identical <div>s in each. AT MOST ONE of the 10 <a> elements is displayed as if the two divs were in separate <a>s, but they're not. The other 9 elements, or sometimes all 10, are displayed correctly. Reloading the exact same file sometimes switches between correct and incorrect display. WHICH ONE of the 10 elements is displayed incorrectly, and in what proportion of successive reloads there is an incorrect element at all, depends on the CSS declaration, the size of the HTML comment with the x's, the use of whitespace in the file, whether or not the file is loaded as a local file or via an Apache server, and the phases of the moon. Reproducible: Sometimes Steps to Reproduce:
Created attachment 177352 [details] Test case This is the test case. Remember that I said that it does not always trigger the bug. Loading it here in BugZilla may not trigger the bug (I can't try that before uploading it). If this does not trigger it, please download the file and place it into the document root of an Apache server running on your localhost. This is assuming the bug does not depend on the humidity of the air you are breathing.
Created attachment 177353 [details] Test case - screenshot *sigh* Of course the attachment didn't trigger the bug, just as I expected. So I'm attaching here a screenshot of what the test case displays like for me, when I view the file through a local installation of Apache. One subtle detail about my original bugreport was inaccurate. I said that the element that is rendered incorrectly renders as two elements. Actually, it renders as five elements, only two of which contain text. You can see the little 2x2-pixel blobs in the screenshot which are empty left-floating divs with a 1px red border.
Created attachment 177354 [details] Test case - screenshot (2) Same test case, same situation, but smaller window size. Notice how the first of the five "wrong" elements is in the first row, the other 4 in the second.
Hallo! This is invalid HTML, you know? <a class='test_case'> <div> Some text </div> <div> Some text </div> </a> (Inline-element a must not contain block-element div) tschüs, jürgen
That it's invalid HTML is not relevant to the bug. If a layout bug shows up with invalid HTML it probably shows up with valid HTML too.
Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b2) Gecko/20050227 Firefox/1.0+ Thank you for the report. I can't reproduce from 1) Bugzilla 2) file 3) local Apache Server Any other hints?
I don't know if this helps, but I've just discovered something interesting. When the bug does occur, and then I use File/Save Page As, and I choose "Web page, complete", then the file is significantly changed in such a way that the deviant display could be regarded as correct. Notice that it still affects only ONE of the ten <a>s-with-<div>s, even though they are all identical. Again, WHICH one of the ten is affected depends, among other things, on the HTML comment in the file as well as the amount of whitespace. The original version is: <a class="test_case"> <div> Some text </div> <div> Some text </div> </a> The version modified by File/Save Page As is: <a class="test_case"> </a><div> <a class="test_case"> Some text </a></div> <a class="test_case"> </a><div> <a class="test_case"> Some text </a></div> <a class="test_case"> </a> Kind of makes sense if it's trying to somehow transform it into valid HTML. Also, it now becomes obvious why there are five floating elements, two of which contain text. But it's still bizarre that it does that to only one of the ten identical elements.
Would be good to have a testcase that reliably shows the problem....
Created attachment 178688 [details] testcase2 Using <script></script> in the divs makes the bug appear reliably. I think this is a duplicate of bug 273543.
Yeah, almost certainly.
Boris is right, but I'm duping this to bug 324875 because that's the bug that's about fixing this issue.